Cognitive decline among older persons is a large and growing public health problem. To date, studies of decline in cognitive function seeking to identify the responsible genetic variants have been largely limited to linkage and candidate gene association studies using the more extreme categorical phenotype of Alzheimer's disease. These approaches have identified several rare and one well-validated common risk allele, the APOE [unreadable]4 allele, but the results have also shown substantial inconsistency. A theme of the proposed work is that much of this inconsistency is attributable to the designs typically used in these studies, and that an approach using the design features of population-based studies can complement existing genetic studies. These features (adequate size, rigorous and prospective designs, and participation from a high proportion of age-eligible residents of a defined geographic area) are especially pertinent to efforts to deconstruct the mixture of genetic and non-genetic factors that is found in complex diseases. The proposed work combines outstanding ability to characterize variation in the human genome at the Broad Institute with the rich characterization of the cognitive decline phenotype from two longitudinal studies: an exploratory cohort from a long-term study of cognitive decline (the Religious Orders Study or ROS) and a confirmatory sample from a rigorous population study (the Chicago Health and Aging Project or CHAP). Advantages of these studies include: (1) existing longitudinal data on decline in cognitive function, (2) availability of DMA samples from most subjects, (3) extensive records of non-genetic covariates and, (4) for the confirmatory cohort, a rigorous base in the general population and an ethnic composition of 50% non- Hispanic black and 50% non-Hispanic white. It is proposed to conduct a genome-wide association scan (550,000 SNPs) in 940 people of European ancestry from the exploratory cohort to identify loci associated with cognitive decline. Using a panel of 1500 SNPs selected according to strength of association with cognitive decline, associations found in the exploratory cohort will be validated in the 1128 people of European ancestry from the confirmatory cohort. Validated alleles will be assessed in the 1128 African Americans from the confirmatory cohort and fine mapping will be conducted for alleles found to be validly associated with cognitive decline among European American or African American subjects.